Amphibole genesis via metasomatic reaction with clinopyroxene in mantle xenoliths from Victoria Land, Antarctica

Abstract

Petrographic and major and trace element geochemical features of clinopyroxene, amphibole and glass from composite xenoliths from Mt. Melbourne and Mt. Overlord (Victoria Land, Antarctica) were investigated. Amphibole disseminated in the peridotite matrix appears to grow around clinopyroxene and spinel, and is often associated with high-TiO₂ silicate glass. Clinopyroxene presents a wide range of compositions, from primary unmetasomatized diopside to high-MgO salite, with REE patterns varying from flat (at Yb_N 5Xchondrite) to slightly enriched [(La/Yb)_N 2.7–4.2] at higher HREE contents (Yb_N 9.3–14.3). Amphibole also occurs in veins which, in few cases, grade into glass-rich veins before vanishing into the peridotite. Glasses are not related to amphibole destabilization; on the contrary, they appear to be strictly related to its formation. No chemical differences were noted between glasses related to disseminated or vein amphiboles. Their geochemical features favour a Na-alkaline silicate melt as the metasomatizing agent. Mass balance calculations were used to model the reactions producing amphibole from primary clinopyroxene, and to highlight the nature of the metasomatic agent/s. Trace element contents of the inferred melt/s are comparable to those of the most undersaturated magma found in the area, suggesting a strong link between metasomatism and the magmatism of the Ross Sea Rift system. This hypothesis is further strengthened by the analogy between trace element patterns of clinopyroxene associated with amphibole (cpx-A) and those of clinopyroxene contaminated by the host basalt (cpxBas). On the basis of the Mg/Fe diffusion model, the difference in mg# between these two clinopyroxenes was used to estimate the timing of the basalt infiltration and amphibole formation. Finally, various models for disseminated and vein amphibole relationships are recalled and their application to Antarctic amphiboles is tested. The petrographic and geochemical features of disseminated amphiboles, in fact, do not support the hypothesis that they may derive from differentiated magma after vein amphiboles have crystallized. Both amphibole types may have formed within a similar time span due to the different magma/wall rock ratios which control the mode of melt migration from porous flow to cracking and fracturing.

Introduction

Mantle xenoliths from Mt. Melbourne and Mt. Overlord (Victoria Land, Antarctica) occur in scoria cones and in basanitic to tephritic lavas. Alkaline to tholeiitic magmatism in this area started in the early Eocene in relation to the Ross Sea Rift system, which is situated along the eastern border of the Transantarctic mountains Fitzgerald et al., 1987, Beccaluva et al., 1991a. The rift system is organized in two main branches: one inland, trending NNW–SSE up to Mt. Overlord, while the other, toward the sea, is directed NE–SW. Petrological features and thermobarometric evolution of the lithospheric mantle from Cape Washington to Mt. Overlord were discussed in Beccaluva et al. (1991b), and more recently by Perinelli and Armienti (2002). However, several peculiar petrographic and geochemical features related to the widespread presence of amphibole remain untreated. In particular, the occurrence (and the strict association in several xenoliths) of amphibole, clinopyroxene and glasses provides a unique opportunity for investigating amphibole formation in relation to metasomatism, eventually leading to the hydration of anhydrous mantle.

It is well-known that amphibole represents a key mineral in mantle paragenesis as it is one of the main hosts for alkalis. Its role is crucial in basalt petrogenesis even at high degrees of partial melting (Beccaluva et al., 1998), as well as in mantle glass genesis via decompression or in situ melting processes Chazot et al., 1996a, Yaxley and Kamenetsky, 1999, Yaxley et al., 1997.

Disseminated amphibole growing around clinopyroxene and/or spinel is commonly observed Francis, 1976, Fabriés et al., 1987, Vaselli et al., 1996, Gregoire et al., 1997, Dawson and Smith, 1988, Bodinier et al., 1990, Zanetti et al., 1996, Woodland et al., 1996, Moine et al., 2001, Ionov et al., 2002, though its genesis is often referred to previous, generally poorly defined, metasomatic event/s Zanetti et al., 1996, Witt-Eickschen et al., 1998. Zanetti et al. (1996) distinguish two kinds of disseminated amphiboles: one older, produced by a previous metasomatism, and one younger, occurring only in the proximity of amphibole veins, related to the infiltration into the wall-rock by the same liquid generating the veins. On the other hand, vein amphiboles are usually considered products of direct crystallization from basanite or alkali basalt magmas in the upper mantle Francis, 1976, O'Reilly, 1987, O'Reilly et al., 1991, Dawson and Smith, 1988, Ionov et al., 1997, Witt-Eickschen et al., 1998, Moine et al., 2001. The residual melt afterward infiltrates the peridotite matrix, creating new amphibole in the form of crystallization or reaction products.

Some inconsistency arises in those models where disseminated amphiboles are (i) older and separated from the magmatic event which created the vein amphibole, even if chemical composition may be quite comparable and, in any case, leaving the petrogenesis of disseminated amphibole unsolved; or (ii) younger and formed by differentiated melts, even if major element contents (especially mg#) are very similar, and melt propagating through fracturing precedes porous flow migration. Thus, notwithstanding the large amount of papers reported above, the chemical and temporal relationships between disseminated and vein amphibole is still a matter of considerable debate. Moreover, to the best of the authors' knowledge, glasses related to amphibole formation have not been reported. Most studies regarding amphibole genesis were, in fact, developed on alpine peridotite massifs Woodland et al., 1996, McPherson et al., 1996, Zanetti et al., 1996 where glass cannot be preserved, or else in mantle xenoliths where amphibole and glass were related to either decompressional (Laurora et al., 2001) or in situ melting Chazot et al., 1996a, Yaxley and Kamenetsky, 1999 of volatile-bearing phases; in both cases amphibole is consumed, and glass plus secondary phases are produced.

As it will be shown below, amphibole in lithospheric mantle from Antarctica is growing at the expense of clinopyroxene and is always associated with glass, thus offering a unique opportunity to study the genetic processes (and metasomatic agent/s) responsible for its formation. To this purpose a painstaking petrographic study was accompanied by a major and trace element microanalytical work mainly addressing clinopyroxene, amphibole (disseminated in the peridotite matrix and in the vein) and glass compositional variations. The effects of host basalt infiltration were also studied in order to isolate host basalt contamination from metasomatic processes. Finally, a comparison between petrological features of the inferred metasomatic agent/s, and the basalts of the Ross Sea Rift system is presented, setting metasomatism and magmatism within a unique framework.

Of particular interest is the paper of Gamble and Kyle (1987), who studied an amphibole-glass-bearing wehrlite from Foster Crater (Antarctica) with textural features identical to those observed in the Mt. Melbourne peridotites; their data and model for amphibole, glass and host magma relationships were, in our opinion, really outstanding for the time. Our investigations however include also lherzolites, thus extending the study to more “common” mantle parageneses. Moreover, the interpretation of Gamble and Kyle (1987) of glasses as simple differentiation products from magmas similar to the host basalts after amphibole crystallization does not agree with our study.

Samples were collected during the Italian Antarctic expedition in the late 1980s. Numbers refer to localities, while letters indicate different xenoliths. Amphibole-bearing peridotites occur mainly in one locality north of Mt. Melbourne and, subordinately, in another west of Mt. Overlord. The studied samples represent all the amphibole-bearing xenoliths found (see also Beccaluva et al., 1991b for further description).